This chapter discusses the utilization of machine learning techniques, particularly artificial neural networks (ANNs), to predict the maximum lateral load of steel plate shear walls (SPSWs) based on their geometric dimensions. The chapter focuses on the design challenges posed by SPSWs and their significance in seismic areas. It explores the advantages of SPSWs, such as their energy dissipation capacity, lateral stiffness, and ductility. The chapter details the research methodology, including data preparation, model architecture, and performance metrics. The authors create a numerical model of an SPSW and use finite element analysis to simulate various samples with different geometric dimensions. The dataset includes 60 unstiffened SPSW samples subjected to uniform lateral loading. An ANN with a hidden layer containing 18 neurons is developed to predict the maximum lateral load based on the input parameters representing the geometric dimensions of the components. The study utilizes metrics such as mean squared error (MSE), root-mean-square error (RMSE), and coefficient of determination (R2) to assess the accuracy of the ANN model's predictions. The results indicate that the ANN with 18 neurons in the hidden layer achieves strong performance, with R2 values of 0.939 for training data and 0.936 for test data. The chapter concludes by demonstrating the convergence between the ANN's predictions and the actual finite element analysis results, showcasing the potential of machine learning in optimizing the design of steel plate shear walls. The research underscores the potential of artificial intelligence, specifically neural networks, in enhancing the accuracy and efficiency of structural design processes, particularly for complex systems like steel plate shear walls.
